Challenges in modelling behaviour


Sebastian Funk (@sbfnk)
https://epiforecasts.io

25 January, 2023

Ferguson, Nature, 2007

  1. Set the baseline and determine the effect of departing from it
  2. Assess how and to what extent behaviour should be modelled explicitly
  3. Determine the minimal level of detail required to model differences in behaviour
  4. Quantify changes in reporting behaviour
  5. Predict the response to interventions and health campaigns
  6. Identify the role of movement and travel
  7. Develop models that can be verified against data from digital sources
  8. Inform real-time data collection
  9. Engage in dialogue across disciplines

The COVID-19 pandemic has generated novel data on behaviour

Gimma et al., PLOS Medicine, 2022

In spite of unprecedented data availability, quantifying relevant behaviours has been challenging during the pandemic

Barnard et al., Nat Commun, 2022

Klepac et al., Epidemics, 2018

We do not have a good handle on the relationship between government action and behaviour

Sharma et al., Nat Commun, 2021

Brooks-Pollock et al., BMC Medicine, 2023

van Kerckhove et al., Am J Epidemiol, 2013

Assuming behaviour does not change is a strong assumption

Statens Serum Institut, 2021

Dönges et al., Frong Phys, 2022

“We still lack a validated theory to describe the feedback loop between behaviours and disease”

Perra, Phys Rep, 2021

Forecasting COVID-19 with behavioural data

Munday et al., medRxiv, 2022

  1. Set the baseline and determine the effect of departing from it
  2. Assess how and to what extent behaviour should be modelled explicitly
  3. Determine the minimal level of detail required to model differences in behaviour
  4. Quantify changes in reporting behaviour
  5. Predict the response to interventions and health campaigns
  6. Identify the role of movement and travel
  7. Develop models that can be verified against data from digital sources
  8. Inform real-time data collection
  9. Engage in dialogue across disciplines